My lesson is that you need to A/B test everything, since the thing that works is never what you think will work. For example, this terrible wall-of-text absolu

My lesson is that you need to A/B test everything, since the thing that works is never what you think will work. For example, this terrible wall-of-text absolutely destroyed a pretty 37 Signals-esque 3-column pricing chart in A/B testing: https://www.s3stat.com/Pricing.aspx https://www.s3stat.com/Pricing.aspx I didn't believe it, so I tweaked it and ran the test again. Exactly the same result. So those paragraphs stay, with the same unordered order, goofy typography and everything else. Nobody touch anything! It's perfectly balanced at its local maxima.